Considering the Complexity of Exchange

We also need to consider the complexity of the exchange between users, users and
devices, and devices. In a voice network, we have said that the exchange is essentially
symmetric and duplex (bidirectional)—a point-to-point exchange.
In a broadcast application, traffic is completely asymmetric. Television would be an
example of an asymmetric point-to-multipoint application. However, we need to consider
that a lot of broadcast content now contains trigger moments—voting in a game
show, for example. Abroadcast might trigger a simultaneous response from all or some
of the audience generating an instantaneous requirement for uplink bandwidth.
Broadcast SMS is an example of a one-to-many exchange that can trigger a many-toone
(multipoint-to-point) response. This defines an important traffic property—the
need for instantaneous uplink bandwidth in response to a downlink broadcast.
This loading phenomenon can be observed in the national electricity grid. At the
end of a popular TV program or in the commercial break, power demand suddenly
surges as everyone goes to put the kettle on to make (in Britain) a pot of tea. The
national grid has to be substantially overprovisioned to support these very concentrated
and sudden loading peaks. Similarly, with gas distribution, when 20 million
households all cook Christmas dinner at the same time, gas pressure drops. Demand
does not go down; ovens have thermostats, so they are smart enough to know they
need constant heat.
Anyway, we digress. The point about broadcast streams or trigger streams is that
they can deliver substantial uplink offered traffic loading (and hence uplink offered
traffic value) but are by nature very peaky in terms of needing to deliver instantaneous
access bandwidth. This can either be provided by dramatically overprovisioning
uplink bandwidth or by supporting uplink demand with a mix of delivery and buffer
bandwidth resource (which implies giving subscribers a variable access delay).